How to Rewrite Your Resume for ATS in 2026

By Rasmus AI for The Resume Code · Published 21 Apr 2026 · Updated 24 Aug 2026 · 12 min read

Direct answer

To rewrite your resume for an ATS in 2026: extract keywords from the exact job description you are applying to, mirror those terms in your summary and bullets, replace duty verbs with measurable achievements, use a single-column layout with standard headings, save as both PDF and DOCX, and validate parsing with a free analyzer like The Resume Code before you submit. Skip headers, footers, tables, and two-column layouts entirely — they are the single most common cause of an otherwise strong candidate getting silently filtered out at the parse step. One correction up front: the ATS is very rarely the thing that rejects you. In a 2026 study of 25 US recruiters, 92% said their systems do not auto-reject resumes for formatting, content, or design. What bad formatting actually does is make you unfindable and unreadable, which produces the same silence by a different route.

Key statistics

  • 97.4% — of Fortune 500 companies had a detectable applicant tracking system in 2026 — 487 of 500 — so your resume is parsed into fields before anyone reads it. (Jobscan ATS usage analysis, 2026)
  • 92% — of recruiters surveyed said their applicant tracking system does not auto-reject resumes for formatting, content, or design; only 8% had auto-rejection configured. (Enhancv survey of 25 US recruiters, 2026)
  • 7.4 sec — median time recruiters spend on the initial scan of a resume, measured by eye tracking — which is the real screen you are formatting for. (Ladders eye-tracking study)
  • 3.5% — of the 1.2 million people who used AI job-search features in 2025 used it to tailor a resume to a specific job description — the step with the most evidence behind it is the least-used one. (Kickresume AI job search data, April 2026)
  • 62% — of employers in Resume Now's AI applicant report said they reject AI-written resumes that lack a personal touch. (Resume Now AI Applicant Report, 2025)

Editor's note

ATS optimization is not stuffing keywords into a footer. It is making sure the most important word in the job posting is the first word of the matching bullet in your resume. Position is signal.

What is an ATS, and what does it really do to your resume?

An applicant tracking system is software that ingests your resume, parses it into structured fields — name, employers, dates, titles, skills, education — and stores it in a searchable database alongside every other applicant. Workday, Greenhouse, Lever, iCIMS, Bullhorn, and Taleo dominate the market. Jobscan's 2026 analysis found a detectable system at 97.4% of Fortune 500 companies — 487 out of 500. Assume you are being parsed.

What the software does next is where most advice goes wrong. It ranks and filters; it very rarely bins. Enhancv interviewed 25 US recruiters across tech, healthcare, and finance in 2026 and found 23 of them — 92% — said their systems do not automatically reject resumes for formatting, content, or design. Only 8% had content-based auto-rejection configured at all. Recruiter after recruiter described filtering to decide who to review first, not who never gets seen.

So where does the widely repeated "75% of resumes are rejected by a robot" figure come from? No credible study. It is a number that has circulated for over a decade without a traceable source, and we have removed it from this guide rather than repeat it. The honest version is less dramatic and more useful: what does reject you is a knockout question you answered wrong, a requirement you do not meet, and volume — a role that pulls a thousand applicants in three days where the recruiter stops once they have a shortlist.

Why formatting still matters enormously: If your resume parses into garbage, you do not get auto-rejected — you get something worse. You land in the database with an empty Skills field and a mangled work history, so you never surface in the recruiter's search, and when a human does open you, the 7.4-second scan finds nothing. Same silence, different mechanism.

Which formatting choices actually break parsing?

Parsers fail predictably. Bullhorn's documentation on resume parsing describes extracting candidate details from the document's text layer and mapping them onto record fields — which tells you what defeats it: content that is not in a linear text layer, or whose reading order does not match its visual order.

Two parallel columns. The parser reads across the page, not down the column, and your skills sidebar interleaves with your job history. • Contact details in a header, footer, or text box. Many parsers ignore those regions, so your phone number and email never make it into the record. • Tables used for layout. Cell boundaries become arbitrary line breaks, and dates detach from their roles. • Text baked into an image or logo. There is no text layer to read, so it does not exist. • Scanned PDFs. Same problem across the whole document; only OCR recovers anything, and most systems do not attempt it. • Non-standard section headings. "My Story" and "Where I've Been" do not map to Summary and Experience, so the content lands nowhere useful. • Custom fonts embedded from a design tool. Substitution can collapse spacing and glue words together.

Common formatting choices ranked by ATS-safety. | Single column, plain headings | Yes | The default and the safest option. | Two-column layout (skills sidebar) | No | Often parsed in the wrong order or merged into one paragraph. | Header with contact in a text box | No | Many parsers skip the header entirely. | Bullets with • or – characters | Yes | Standard bullet characters parse cleanly. | Custom fonts loaded from a design tool | No | Substitution can collapse spacing and break parsing. | Dates as "Mar 2023 – Present" | Yes | Unambiguous month-year ranges map to date fields reliably. | Skills shown as bars or star ratings | No | Graphics carry no text, so the skill is invisible to the parser.

What does an ATS-friendly format actually look like?

Single-column layout, top to bottom. No sidebars. • Standard fonts: Calibri, Arial, Helvetica, Source Sans, Inter, or Garamond. 10.5–11.5pt body, 14–18pt name. • Standard section headings: Summary, Experience, Education, Skills, Projects, Certifications. Avoid creative replacements like "My Story". • Contact info in plain text, not in a header, footer, or text box. • No tables, no graphics, no icons embedded as images. • Month and year for every date range, written the same way throughout. • Save as PDF for the human reviewer and DOCX in case the ATS requires it. Test both.

None of this is a design compromise so much as a discipline. A single-column resume with generous whitespace and consistent alignment reads better in 7.4 seconds than a two-column layout does — the eye-tracking research shows recruiters scanning the left edge for names, titles, and dates. The format that survives the parser survives the scan.

How do you do keyword research for a single job posting?

Copy the entire job description into a text editor. • Highlight every noun phrase that appears more than once. These are your priority keywords. • Identify required vs preferred skills and tools. Required terms must appear in your resume; preferred terms should appear if true. • Note the exact phrasing — "customer success management" and "customer success manager" are different tokens to many parsers. • Check the knockout questions on the application form itself. These are the filters that genuinely do reject people, and they are usually about eligibility, licensing, or years of experience rather than wording. • Cross-reference with two more postings for similar roles to build a master keyword list of 12–15 terms.

Then place them deliberately. A keyword buried in the sixth bullet of a job you left in 2019 is technically present and practically invisible. The terms that matter most belong in your summary, your Skills line, and the first bullet of your most recent role — the three places both a keyword search and a seven-second human scan actually land.

Do not lie: Only mirror keywords you can defend in an interview. Inflated skills get caught at the screening call and burn the relationship for life — and 65% of HR leaders already say AI-enhanced resumes make candidate skills harder to verify, so scrutiny is rising, not falling.

How do you rewrite a bullet to match a job description?

The single most important rewrite move is replacing duty verbs with achievement verbs and quantifying the result. Here is a worked example for a software engineer applying to a backend role that emphasizes performance and reliability.

Before

Responsible for working on the payments service and helping with on-call rotation.

After (rewritten by Rasmus AI)

Reduced p99 latency on the payments service from 480ms to 110ms by introducing a Redis-backed cache and rewriting the settlement query, while owning a weekly on-call rotation that closed 38 incident tickets in six months.

The pattern: Action verb → measurable outcome → method → context. Every bullet that follows this pattern out-competes the same content written as a duty.

The same move outside engineering

Duty: "Managed the company's social media accounts." → Achievement: "Grew organic LinkedIn following from 4,200 to 19,000 in eleven months by shifting to a three-post-per-week founder-voice cadence, lifting inbound demo requests 34%." • Duty: "Handled accounts payable." → Achievement: "Cut invoice processing time from nine days to two by consolidating three approval queues into one workflow, eliminating £41k in annual late-payment penalties." • Duty: "Supported nursing staff on the ward." → Achievement: "Coordinated discharge planning for a 28-bed ward, reducing average discharge delay by 4.5 hours and freeing an estimated 300 bed-hours per quarter."

If you genuinely do not have a number, use scale, frequency, or scope instead: how many people, how often, how large the budget, how many systems. "Ran the weekly release for a team of nine across three services" is specific even without a percentage. What you should never do is invent one — 94% of hiring managers report having already encountered misleading or inaccurate AI-generated content in applications, and a fabricated metric that survives to interview ends the conversation permanently.

How do you validate that the rewrite actually parses correctly?

Before you submit, run the new file through The Resume Code's free analyzer. We extract the text from your uploaded file and check what came through: standard headings, a findable email and phone number, table-like structures, and keyword coverage. That is not an emulation of any employer's ATS, and it misses page geometry — fonts, text boxes, and column order are invisible to any text-level check. So do the manual test too: open the PDF, select all the text, copy it into a plain-text file. If the order is jumbled or sections are missing, the ATS sees the same garbage.

Copy-paste the whole document into a plain-text file and read it top to bottom. Reading order problems show up immediately. • Check that your name, phone, and email survived. If they came from a header or text box, they usually did not. • Check that every employer sits next to its own dates and title. • Confirm your Skills section came through as readable words rather than disappearing with the graphics. • Repeat the test on the DOCX version, not just the PDF — they can parse differently. • Re-run the check after your final round of manual edits, which is where formatting damage usually sneaks back in.

Once the file passes parsing, do one more pass for human readability. The ATS gets you to the queue. The recruiter still has to want to call you.

How much should you tailor, and how often?

Tailoring is the step with the strongest evidence behind it and the lowest adoption. Of the 1.2 million people Kickresume observed using AI job-search features in 2025, about 64% used it to check a resume against an ATS, roughly half used it to write or improve content — and only about 3.5% used it to tailor a resume to a specific job description. Almost everyone is validating; almost nobody is targeting.

You do not need a new document per application. Maintain one master resume with everything on it, then per posting rewrite the summary, the Skills line, and the top three bullets of your most recent role. That is roughly fifteen minutes of work once the master exists, and it is the portion of the document that both keyword search and human scanning actually weight.

Rasmus AI: The candidates who get interviews are not the ones who found a magic template. They are the ones whose first three bullets read like they were written for that specific job, because they were. Everything below that is supporting evidence.

Frequently asked questions

Do applicant tracking systems really reject 75% of resumes automatically?
No — that figure has no credible source and we no longer cite it. In a 2026 survey of 25 US recruiters, 92% said their systems do not auto-reject on formatting, content, or design, and only 8% had auto-rejection configured. What does filter you out is a knockout question, an unmet requirement, or simply volume.
If the ATS doesn't reject me, why does formatting matter?
Because bad formatting makes you unfindable rather than rejected. A resume that parses badly lands in the database with empty fields, so you never appear in the recruiter's search — and when a human does open it, the roughly 7.4-second initial scan finds nothing to hold on to.
Will a one-page resume always parse better than a two-page resume?
No. Length is not what kills parsing. Layout, fonts, tables, and text boxes are. A clean two-page resume parses fine; a designer-built one-page resume often does not.
Should I include a Skills section even if the same skills appear in my bullets?
Yes. Many ATS engines weight the Skills section more heavily than narrative bullets, and recruiters searching the database often search on exactly those terms. Repeat your most important skills there, separated by commas or pipes — as text, never as bars or star ratings.
Do I need to write a different resume for every job?
Tailor the summary, the Skills line, and the top three bullets of your most recent role to each posting. The rest can be a stable master resume you maintain over time. Only about 3.5% of AI-assisted job seekers tailor at all, so this is where the remaining advantage sits.
Are PDFs or Word documents safer for ATS submission?
Modern ATS engines parse both well. Submit the format the application asks for. If it does not specify, prefer PDF for layout fidelity. Keep a DOCX version on hand and test that it parses too — the two can behave differently.
Should I use ChatGPT to rewrite my resume?
ChatGPT can help phrase individual bullets, but it never sees your actual file and does not score against a rubric. Pair it with a dedicated tool that extracts the text from your document and checks what survived — and always edit the output into your own voice before submitting.
What is the fastest single change I can make today?
Rewrite the top three bullets of your current job using the action-verb-then-measurable-result pattern. That single change moves more interviews than any other one-hour edit.

Sources